uMR Omega with uWS-MR-MRS

The uMR Omega with uWS-MR-MRS introduces significant enhancements to a 3.0T magnetic resonance imaging system, aiming to improve diagnostic image quality and expand clinical applications. This is achieved through the incorporation of new multi-element RF coils, novel pulse sequences, advanced AI-driven image reconstruction (DeepRecon), expanded quantitative and spectroscopic capabilities for various organs, and new functionalities designed to improve workflow and patient safety.

The uMR Omega is a 3.0 Tesla superconducting magnetic resonance diagnostic device (MRDD) featuring a 75cm patient bore, designed for producing cross-sectional and spectroscopic images of the head, body, and extremities to aid in diagnosis. It utilizes components like a magnet, RF power amplifier, various RF coils, gradient systems, and a spectrometer, conforming to NEMA and DICOM standards. The accompanying uWS-MR software is a comprehensive solution for processing, reviewing, analyzing, and storing MR studies, with applications including image stitching, dynamic studies, image fusion, MR spectroscopy (SVS and CSI), quantitative mapping, breast evaluation, brain perfusion, vessel analysis, virtual endoscopy (Inner view), DCE analysis, neuro imaging, and cardiac analysis. These tools aim to provide detailed anatomical and functional information, with contrast agents usable depending on the scan's region of interest. This submission details modifications primarily to the uMR Omega and its associated uWS-MR software, including the addition of several new Radio Frequency (RF) coils (e.g., Head & Neck Coil-48, Spine Coil-48, Head Coil-64, SuperFlex Body-24), new and modified pulse sequences (e.g., gre_fact, asl_3d, fse_mars_sle, T1/T2/T2* mapping), and advanced image processing methods like Fat Analysis and Calculation Technique (FACT), Arterial Spin Labeling (ASL), Flow Quantification (FQ), and DeepRecon, an AI-based image de-noising and super-resolution algorithm trained on over 165,000 cases and validated on diverse US subjects. Spectroscopy capabilities have been expanded to include liver, prostate, and breast applications. New functionalities such as Remote Assistance, MR conditional implant mode, and expanded EasyScan support for additional body parts have also been introduced. Performance data, including non-clinical testing and evaluations by American Board of Radiologists certified physicians, confirmed that DeepRecon images achieved higher SNR (302.48 vs 209.41 without DeepRecon) with minimal impact on uniformity (0.15% difference) and contrast (0.9% difference), and maintained structural measurement accuracy, verifying that these modifications meet clinical diagnostic requirements.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the performance of the DeepRecon algorithm, particularly in terms of noise reduction and super-resolution, compare quantitatively against established non-AI based reconstruction techniques across different anatomical regions and pathologies?; What are the specific improvements in signal-to-noise ratio (SNR) and image acquisition time achieved with the new RF coils (e.g., Head Coil-64, SuperFlex Body-24) compared to their predecessors for their intended applications?; For the newly added spectroscopy sequences for liver, prostate, and breast, what is the demonstrated diagnostic accuracy, sensitivity, and specificity in detecting characteristic metabolites relevant to specific pathologies in these organs?; Could you elaborate on the validation process for the "MR conditional implant mode," detailing the range of implant conditions tested and the system's response mechanisms to ensure patient safety?; What impact do the new pulse sequences (e.g., gre_senc_spiral, asl_3d for CBF imaging) and imaging processing methods (e.g., Flow Quantification) have on quantitative biomarker accuracy and their utility in specific clinical diagnostic scenarios?